WIXLIB

2DWDM Fundamentals, Components, and Applicationswhen the optical/electrical/optical conversion would be unnecessary andquestionable.Today s telecommunications networks have benefited from research insymbiosis in the fields of materials, electromechanics, microelectronics, computing, and optics.Nowadays, low nonlinearity fibers, dispersion management on fibers,lasers with high wavelength stability over years, new tunable light sources,new erbium-doped amplifiers, distributed Raman amplifiers, semiconductoramplifiers and gates, wavelength converters, couplers, space-switches, andwavelength division multiplexers offer new solutions. DWDM with morethan 160 channels, OADM with more than 32 channels, and wavelengthselective cross-connects (WSXC) with 32 32 input-to-output ports are currently in commercial use. Wavelength-interchanging cross-connects (WIXC)will become available soon, as the first industrial tunable lasers and wavelength converters are already proposed on the market.We did our best to organize the text of this book into a logical sequenceof chapters. After the short introduction of this chapter, the basic principlesand background in wavelength division multiplexing are reviewed inChapter 2. Chapter 2 also includes an introduction to soliton multiplexing,one of the options for long-distance transmissions. In Chapter 3, we reviewthe current choices available in DWDM passive components. In Chapter 4we present the main active sources including semiconductors lasers, glassdoped lasers, broadband sources for spectral slicing, and the wavelength conversion technology. In Chapter 5 we review the different solutions in opticalamplification. Optical switches, routers, cross-connects, and add/drops aredescribed in Chapter 6, after discussions on topological configurations andarchitectures of optical networks in which they are utilized.In Chapter 7, the optical fiber nonlinearities limiting the performanceof DWDM systems are presented. Chapter 8 gives examples of applicationsof the DWDM technology and results in long-haul transmission networksand in different passive or wavelength-switched voice and data networksincluding the Internet.

2Basic Principles and Background2.1 Wavelength Division Multiplexing: Basic PrinciplesTelecommunications makes wide use of optical techniques in which thecarrier wave belongs to the classical optical domain. The wave modulationallows transmission of analog or digital signals up to a few gigahertz or gigabits per second on a carrier of very high frequency, typically 186 to 196 THz.In fact, the bit rate can be increased further, using several carrier waves thatare propagating without significant interaction on the same fiber. It is obvious that each frequency corresponds to a different wavelength. This technique is called frequency division multiplexing (FDM) or wavelengthdivision multiplexing (WDM). The latter term is currently preferred in mostcases. DWDM is reserved for very close frequency spacing (typically less than100 GHz corresponding to 0.8 nm at wavelengths near 1.5 mm). The term frequency division multiplexing is used in a few cases, such as multiplexingwith optical frequency shift keying and coherent detection. But the terminology is not completely stabilized.With WDM, it is possible to couple sources emitting at different wavelengths l1, l2, lj, ln into the same optical fiber. After transmission onthe fiber, the l1, l2, ln signals can be separated towards different detectors at the fiber extremity (Figure 2.1). The component at the entrance mustinject the signals coming from the different sources into the fiber with minimum losses: This is the multiplexer. The component separating the wavelengths is the demultiplexer. A simple optical coupler may replace themultiplexer, but losses will increase. Obviously, when the light propagation3

4DWDM Fundamentals, Components, and Applicationsl1 l2 l 3 l i l jljljlil1l2l3lil3l2l1Figure 2.1 WDM.is reversed, the multiplexer becomes the demultiplexer, and the reverse is alsotrue. It is important to note, however, that the coupling efficiency is not necessarily preserved in reverse operation. For example, if the multiplexer usessingle-mode (SM) entrance fibers and a multimode output fiber, the coupling losses would be excessive in the reversed usage. Multiplexers designedwith identical input and output fibers are usually reversible. Simultaneousmultiplexing of input channels and demultiplexing of output channels canbe performed by the same component, the multi/demultiplexer.2.2 History of WDM in a Few WordsThe optical multiplexing concept is not new. To our knowledge, it datesback to at least 1958 [1, 2]. Perhaps we can say that the idea of sendingmultiple signals, as shown in Figure 2.1, was straightforward, as it was atransposition of techniques used in classical telecommunications with electronics signals. But the technical problems to be solved were very difficult,and it took experts a great deal of time to solve them. About 20 years later,the first practical components for multiplexing were proposed primarily inthe United States, Japan, and Europe. In 1977, the first grating-WDM passive component was developed by Tomlinson and Aumiller [3].2.3 WDM and Time Division MultiplexingIs it easier to multiplex the signal in the electronic domain time divisionmultiplexing (TDM), or in the optical domain (FDM or WDM)? The

Basic Principles and Background5answer to this question is not easy and the optimum solution is generallyfound in the association of the different techniques.For low bit-rate services ( 2 Mbps), it is generally better to use onlyTDM techniques. For uncompressed, high-definition television (HDTV)broadcasting, WDM is highly recommended. The video compression techniques minimize the bandwidth requirement. However, at the time of thiswriting, CATV and HDTV still require 4 Mbps and 25 Mbps, respectively.Applications such as video networks linking workstations, television studiocenter signal routing systems, video conference networks, interactive videotraining systems, bank information service networks, and data-transfer networks between computers, integrated service digital networks (ISDN), teledistribution, and generally all broadband networks increasingly use bothtime and wavelength multiplexed optical lines. Today, the predicted demandper subscriber in 2010 is on the order of 100 Mbps. That will not be possiblewithout the deployment of DWDM optical fiber networks.It is understood that a practical network is very often made up of anassociation of architectures that constitute the physical medium of the network between stations. The topology is called virtual when it is concernedonly with logical connections between stations. One example of an opticalmultiplexing application is to create virtual topologies on request. The network configuration can be modified independently of its physical topologyby changing the emitted or received optical frequencies. In these architectures WDM cross-connectors, WDM routers, and WDM add/drops becomemore and more important.2.4 Wavelength Domain and Separation Between ChannelsWith modern, commercially available telecommunication fibers, it is possible to transmit information over a large spectral range (Figure 2.2) with twodomains with low attenuation, one around 1.3 mm and another around1.55 mm. Between these two domains there is generally a high attenuationat 1.39 mm due to the residual OH radical in the fiber. SM silica fibers withminimum loss of about 0.16 dB/km at 1.55 mm are available. Losses 0.4dB/Km at 1.5 mm and 0.5 dB/Km at 1.31 mm are specified in the ITU-TG.652 recommendations. In medium- and long-distance DWDM transmissions, SM fibers are generally used in the domain 1,520 to 1,620 nm (seeparagraph 5), due to the availability of efficient optical amplifiers andsources.

6DWDM Fundamentals, Components, and ApplicationsdB/km1.91.390.6 mm105Absorption peak of OHWithout OH150300THznFigure 2.2 Typical loss of a low OH content fiber.For applications such as metropolitan area networks, special SM fiberswith very low OH content are commercialized. They can be used from 1,335to 1,625 nm (for instance, AllWave fiber from Lucent Technologies).On multimode silica vapor phase axial deposition (VAD) fiber withoutOH impurity, losses lower than 4 dB on 2.4 km were obtained from 0.65 to1.9 mm as long ago as 1982 [4].On multimode fibers, a graded index design allows a temporal dispersion minimization, but the optimum profile depends greatly on wavelengthand material, and the dispersion varies with wavelength.The separation between channels is now 0.8 nm or more on mostof the installed networks using WDM. International TelecommunicationUnion (ITU) standardization proposes a frequency grid with separations of100 GHz (about 0.8 nm) with multiples and submultiples. Now the published minimum channel spacing is about 0.1 nm. However, at the beginning of the twenty-first century, nothing lower than 0.2-nm (25 GHz)spacing was commercially available. At first glance, a fiber without OHwould allow 1,000 channels at 50-GHz spacing to be multiplexed over itslarge spectral range!Of course there are some limitations of WDM (Figure 2.3). The mainproblem is crosstalk (parasitic light) coming from technical defects in thedemultiplexers, but also from physical problems such as wavelength conversion along the transmission fiber by four-wave mixing, Brillouin or Ramaneffect, or other nonlinear effects. But the theoretical minimum channel spacing is at last related to uncertainty relationships.

Basic Principles and Background7l1l2 From sourcesl3l1 l 2 l 3To fiberMultiplexerFigure 2.3 Principle of multiplexing by diffraction on an optical grating: Wavelengths l1,l2, l3 coming from different directions are diffracted in the same directioninto a single-transmission fiber.2.5 Wavelength AllocationThus far, the ITU-T standards recommends 81 channels in the C band witha constant spacing of 50 GHz anchored at 193.1 THz. This range can beextended to the L band (191.4 to 185.9 THz) where sources and amplifiersbecome available now. This will add 111 channels at 50-GHz spacing (seeTable 2.1).2.6 Optical Wavelength/Optical Frequency ConversionDepending on one s background (the classical optical field or the microwavefield), one generally prefers to represent light vibrations according to theirwavelengths in vacuum (the wavelength varies with the medium) or according to their frequencies (invariant of the medium). In order to evaluate thedifferent results given in scientific papers, it is useful to be able to translatequickly.It is well known that:l cnwhere c is the speed of light (c 2.9972458 106 m/s), l is the wavelength invacuum, and n is the optical frequency.

8DWDM Fundamentals, Components, and ApplicationsTable 2.1Frequency Standards with Corresponding WavelengthsC Band(196.1 192.1 THz)L Band(191.4 185.9 THz)Frequency(THz)Wavelength (Vacuum)(nm)Frequency(THz)Wavelength 3.41,550.12188.71,588.73

Basic Principles and BackgroundTable 2.1 (continued)C Band(196.1 192.1 THz)L Band(191.4 185.9 THz)Frequency(THz)Wavelength (Vacuum)(nm)Frequency(THz)Wavelength 559.79187.51,598.89192.11,560.61187.41,599.75 187.31,600.6 187.21,601.46 187.11,602.31 1871,603.17 186.91,604.03 186.81,604.88 186.71,605.74 186.61,606.6 186.51,607.47 186.41,608.33 186.31,609.19 186.21,610.06 186.11,610.92 1861,611.79 185.91,612.659

10DWDM Fundamentals, Components, and ApplicationsTherefore, l l2 ncIn practice,l 299,792.458nwith l in nm and n in GHz l nm n GHz l2mm0.310 3Thus, 100-GHz spacing at 1.55-mm wavelength is equivalent to a spacing of0.8 nm.2.7 How Many Channels?There is actually about 15,000 GHz of optical frequency bandwidth in each1,300- and 1,550-nm window. With a 10 Gbps bit rate, the uncertainty relationship gives approximately 10 GHz as the limit for optical frequency spacing. This would mean 1,500 channels! But the fiber nonlinearity has alwaysset the limit to a few hundred channels in practical applications.For the component itself at this limit, the optical crosstalk is the mainproblem. Acceptable 160-channel grating-WDM components are alreadymanufactured. We believe that many more channels are feasible. Threethousand-channel classical-grating spectrometers are commonly used, whynot WDM networks with a few hundred channels? In fact, the main problemis acquiring enough stable fixed or tunable sources. Today, the practical limitis a few tens of sources spaced at 50 GHz. The number of channels willdepend on the progress on the sources. It is worth pointing out that an optical frequency chain-generation from a single supercontinuum source withover 1,000 channels at 12.5-GHz spacing has been proposed by H. Takara,et al. in ECOC 2000 (see Chapter 4).

Basic Principles and Background112.8 Some Definitions2.8.1LossesThe multiplexer must combine the signals with minimal losses. Those lossesPj are expressed in decibels (dB) at each wavelength lj by: Φj P j 10 log Φ0 where Fj is the optical power injected into the transmission line and F0 is theincident power at lj.2.8.2CrosstalkAt the other end of the fiber, the signals at the different wavelengths areseparated by a demultiplexer which, like the multiplexer, must have minimallosses. The optical crosstalk Dij of a channel i on a channel j is: Φij Dij 10log Φjj where Fij is the residual optical power of channel i at wavelength li in channel j and Fjj the exit optical power in channel j at wavelength λj.The total optical crosstalk in channel j is: Φij i j D j 10log Φjj This defect is merely due to the demultiplexer when sources with spectralwidths much smaller than the multiplexer spectral passbands are used. But italso becomes necessary to take into account the multiplexer crosstalk in othercases. In such cases (such as LED slicing) the optical crosstalk is a complicated function of sources, multiplexer and demultiplexer.The electrical crosstalk also depends on the receivers. At equivalentreceiver sensitivity, in general, electrical crosstalk is twice as small in decibelsas the optical crosstalk. The electrical crosstalk of a system also depends onthe relative power and spectral width of the emitters, on the fiber spectral

12DWDM Fundamentals, Components, and Applicationstransmission, and on the receiver s sensitivity variation with wavelength. Forexample, the difference is about 50 dB in a silicon receiver between 0.8 mm(high sensitivity) and 1.3 or 1.5 mm (almost no sensitivity).When the same optical component performs multiplexing of somewavelengths and demultiplexing of the other wavelengths, we use a similarcrosstalk definition, but the term near-end crosstalk is used for the parasiticeffect of sources on receivers in the vicinity of these sources. The crosstalkcoming from sources located at the other end of the line is called far-endcrosstalk. It is obvious that the intrinsic near-end crosstalk of a componentmust be several orders of magnitude lower than its far-end crosstalk specification because the noise of sources not attenuated by the transmission line issuperimposed through the component crosstalk to the signal of a sourceattenuated in the transmission line. Components with intrinsic near-endoptical crosstalk lower than 120 dB have been manufactured [5]. Thenear-end crosstalk also depends on the different reflections along the line, inparticular on the number of connectors and on their location along the transmission line.2.9 Solitons2.9.1Soliton PropagationThe pulse broadening induced by chromatic dispersion can be compensatedfor by a self-phase shift induced by refraction index variations upon the fieldintensity (known as the Kerr effect) for particular pulses that have a criticalpower and shape called solitons. With silica glass the nonlinear index isn n0 3.2 10 16 I, with I in watt/cm2 and in which n0 is the index at anarbitrary low-intensity I. The response time of this phenomenon is arisingfrom electronic structure distortion in less than 1 fs (10 15s 1 fs) and fromnuclei motion in about 100 fs in silica.In soliton propagation, the velocity of the trailing half of the pulsetends to be increased and the velocity of the leading half of the pulse tends tobe decreased by the nonlinear effect; this compensates for the reverse effectdue to f

3.3.5 Typical Specifications of Available Bragg Grating DWDM 31 3.4 Optical Multidielectric Filters 32 3.4.1 General Principles 32 3.4.2 Materials and Processes for Enhanced Performances 37